Installation and Downloads
cim-optimizer requires Python version 3.7, 3.8, 3.9, or 3.10. Installation of cim-optimizer can be done via pip:
pip install cim-optimizer
Compiling from source
cim-optimizer has the following dependencies:
Python >= 3.7
NumPy >= 1.19.2 (>= 1.22.3 for GitHub Actions pytests)
matplotlib >= 3.5.2
Torch >= 1.12.1 (and additionally CUDA 11.6 for GPU acceleration)
To compile the latest developments, clone the git repository and install using pip in development mode:
git clone https://github.com/mcmahon-lab/cim-optimizer.git
cd cim_optimizer && python -m install -e .
PyTorch Support
To use cim-optimizer with PyTorch using CPU, it is sufficient to run
pip install torch
However, if GPU acceleration is desired, it is expected to instead run
pip3 install torch --extra-index-url https://download.pytorch.org/whl/cu116
Please refer to PyTorch‘s website for additional information.
Software Tests
To confirm the accuracy of cim-optimizer’s performance, users can locally run the test suite via pytest. To perform this, pytest must be installed via:
pip install pytest